Balancing personalization and risk sharing: the future of the Belgian non-life Insurance Industry in the Age of AI

The balance between individualization and risk sharing

For insurance products where risk is influenced by consumer behavior, the era of standardized packages is fading, replaced by tailored products that let customers choose coverage options based on their specific circumstances.

For instance, usage-based car insurance where premiums are calculated based on actual driving behavior or distance driven (also known as Pay as/how you drive: PAYD & PHOD), is gaining traction in the auto insurance market. Popularized in Belgium in the early 2000s, the concept of usage-based insurance has grown significantly. And in Europe, approximately 20% of car insurance policies were based on usage-based insurance by 2023. This figure has grown due to consumer demand for more personalized insurance options (source : Global Market Insights Inc.). Similarly, insurance providers are offering personalized healthcare programs, both physical and mental, to individuals and companies. These programs not only help customers manage their health but also incentivize behaviors that reduce long-term healthcare costs and absenteeism for employers. By personalizing products in such ways, insurers can position themselves as trusted partners in the customer’s journey, fostering greater loyalty and trust.

However, this increasing personalization poses a challenge for insurers. While customers desire customized insurance products that reflect their individual risk, this approach conflicts with the traditional model of risk pooling, where premiums are based on the collective risk shared across a larger group. The principle of insurance relies on balancing individual risks within this pool, with premiums calculated for different levels of risk to ensure fairness and sustainability. In some types of insurance, such as fire insurance, where risk is not behavior-dependent and relevant data is harder to gather, personalization offers limited value. Here, insurers still depend on conventional methods to assess and pool risk, as behavioral data plays a minimal role in determining premiums. Maintaining the equilibrium between individualized pricing and the broader risk-sharing model is critical to the long-term viability of insurance.

AI and advanced analytics in Insurance

As personalization and data-driven assessments become more precise, AI and advanced analytics are reshaping the competitive landscape among insurers. Insurers that effectively leverage these tools gain a significant advantage in both pricing and risk selection, allowing them to attract, select, and retain good-risk profiles—customers with lower risk—at competitive rates. This approach makes these individuals more profitable for insurers.

However, these personalized models pose challenges for customers with bad-risk profiles, who may face substantially higher premiums. While data-driven insurers can use AI not only for pricing but also for risk acceptance decisions, higher-risk individuals might turn to traditional insurers. The latter lack cutting-edge data analytics and offer less personalized coverage at less expensive rates compared to AI-driven alternatives.This shift creates a potential divide in the market, where data-driven insurers secure the most profitable customers, while traditional insurers take on the higher-risk segment, potentially destabilizing the traditional risk-pooling model at a broader scale.

Ethical risks in AI and data analytics

At the core of this customer-centric transformation lies technology. While AI and Data analytics offer significant opportunities for insurers to better understand and meet the needs of their customers, they also introduce complex ethical challenges. As mentioned above, the rise of highly personalized pricing models also raises concerns about affordability, particularly for those deemed high-risk.

The European Insurance and Occupational Pensions Authority (EIOPA) has been proactive in developing ethical frameworks to address these risks. One of the main concerns is algorithmic bias, where AI-based pricing and risk assessment algorithms could inadvertently disadvantage certain customer groups based on incomplete or biased data sets. This could lead to unfair treatment, where individuals are charged higher premiums or denied coverage due to factors beyond their control. To address this, insurers may need to decide between augmented underwriting, where human judgment is combined with AI insights to ensure fairness, and algorithmic underwriting, which relies solely on automated processes.

To mitigate these risks, the "Human-in-the-Loop" principle plays a crucial role, ensuring that human oversight remains a key part of AI decision-making processes. By involving human experts in critical stages of AI-driven decisions, insurers can ensure that the final outcomes are fair, transparent, and ethical, helping to prevent biases.

Privacy concerns are also paramount. Advanced technologies, such as behavioral insurance and AI-driven products, rely on monitoring customers' actions, from driving habits to lifestyle choices. This can raise serious concerns about privacy. Even when customers give their consent for data collection, it does not necessarily eliminate the discomfort people might feel when all their actions are tracked, and this ongoing tension could undermine trust in insurance providers.

Regulatory efforts, such as the European Commission’s Financial Data Access (FIDA) proposal, that was introduced in June 2023, are crucial in - among other things - establishing responsible frameworks for managing customer data in non-life insurance products while protecting consumers from privacy violations and financial exclusion. Life insurance products have been deliberately excluded, as the draft proposal highlights concerns that data sharing in this area could lead to significant risks of financial exclusion, especially if it results in overly strict or discriminatory underwriting practices.  The unintended consequence of such measures would be the creation of barriers that prevent certain individuals from accessing essential coverage—an outcome no one wants to see materialize.